Research in our lab focuses on collecting and analyzing high-density, ecologically valid datasets from parents and children as they interact naturally and in controlled settings.
We are interested in children’s early social-emotional development and maternal mental health and well-being. We use sensors worn by mothers and their children to capture the daily behaviors hypothesized to contribute to these outcomes. For example, to learn more about patterns of children’s distress and soothing, infants and toddlers in our studies wear recorders that capture up to 72 hours of audio data over the course of a week. We can combine this data with information about maternal mood, collected via daily surveys, or mother and infant sleep, collected via devices similar to a Fitbit.
We use these innovative “in the wild” data to characterize daily risk and protective behaviors during the first few years of a child’s life. This time is important because it in these years that patterns of children’s emotion regulation are becoming established. Additionally, the first years of a child’s life also corresponds to the highest risk for mothers’ mental health. Ultimately, we plan to develop interventions embedded in daily life to modify behavioral risks for adverse outcomes – for children and their mothers – while such risks are particularly malleable. We have recently published a paper on the promise of such mobile sensing techniques for developmental science.
Learn more about our research areas below.
Developing a sensor paradigm for developmental science.
A main research effort of the lab has been developing and refining a novel sensor paradigm and pipelines to collect, process, and analyze multimodal datasets of 72 hours of natural mother-child activity data.
Daily activity sensing.
While sleep deprivation and incessant crying are taken as a given of early childrearing, very few studies have documented caregiving experiences with objective data. Using our multimodal sensor paradigm, we aim to objectively characterize the daily behaviors and interactions of mothers and children in the home and study their impacts across the first years of life. For example,
- Do daily sleep and child crying levels predict changes in maternal mood and mental health symptoms?
- How do daily changes in a mother’s mood and social support affect her caregiving?
- Do physical touch and holding affect the development of infant physiological regulation?
- Do patterns of daily motion represent relatively stable individual differences in children’s temperament? Are such individual differences affected by caregiving, and vice versa?
Real-time dynamics of regulation.
The first few years of life have lasting implications for a child’s ability to regulate distress, widely considered a foundation for lifelong mental health. Across studies, child regulation behaviors are strongly related to parenting behaviors. However, much is still unknown about the processes by which such individual differences get established. We hope to gain insight into the behavioral mechanisms that shape these outcomes by studying the real-time dynamics of parent-child distress and regulation. For example, in recent work (in press) we analyze an existing lab freeplay dataset of N=200 depressed mothers and their infants. Our micro-analyses indicate that more depressed mothers – and those with highly negative infants – contingently respond to their infants’ distress at similar rates as other mothers. However, more negative infants benefit much less from such attempts to soothe them, highlighting the importance of infant characteristics to developing self-regulation.
Methods for computational behavioral science.
To make the collection and analysis of high-density behavioral data more accessible, we have organized workshops and published papers for the analysis of multimodal timeseries data (de Barbaro, 2013; Xu, de Barbaro, Abney, & Cox, 2020). We are currently working on a methods paper that will detail our protocol for collecting 72 hours of multimodal audio, motion, physio, and parent-child activity data.
We also aim to develop new ways to objectively characterize developmentally meaningful activity at home from raw sensor data. For example, we have developed models to automatically detect parent holding and touching behaviors from motion (accelerometer) sensors worn by parents and their infants.
Locate our publications via:
Google Scholar (includes publications and other scholarly work);
Dr. de Barbaro’s Google Scholar profile;
PubMed (includes many but not all publications).
Find our NIH grants via NIHRePORTER.